The Edinburgh Disfluency Group Researching disfluency from a psycholinguistic perspective: Language and speech encoding – Grammar – Phonology – Phonetics A general interest in – Speech errors – Speech-error repair and avoidance mechanisms
The Inner speech of People who Stutter - does it contain more errors? Paul Brocklehurst & Martin Corley University of Edinburgh 2010 Funded by The Economic and Social Research Council
The Covert Repair Hypothesis Postma & Kolk (1993) Disfluencies arise because speakers try to repair errors internally, before starting to speak Stutterers are particularly disfluent because their phonological encoding abilities are impaired – they make (and covertly repair) many phonological encoding errors – Their covert repairs also contain errors… sparking off more repairs
The Covert Repair Hypothesis Postma & Kolk (1993) From a listener’s perspective, many covert repairs sound like stuttering-like disfluencies. – E.g. he asked Roger… Robert for dinnerslow detection he asked Ro … Robert for dinner faster detection he asked R … Robert for dinner even faster detection he asked... Robert for dinner fastest of all detection
Stuttering phenomenology – Stutterers do not generally report experiencing problems with inner-speech Picture copied from:
Previous research Few studies have investigated self-reports of speech errors Postma & Kolk (1992) PWS and controls Strings of 4 syllables - spoken out loud - with and without auditory masking Participants pressed a button each time they made an error Findings: In both the normal speech and the noise masked conditions... “No significant group effects were found for the error percentages... or error detection accuracy”
Previous research Few studies have investigated self-reports of speech errors Postma & Kolk (1992) 18 Stutterers and 18 controls Strings of 4 syllables - spoken out loud - with and without auditory masking Participants pressed a button each time they made an error Findings: In both the normal speech and the noise masked conditions... “No significant group effects were found for the error percentages... or error detection accuracy”
The current study Compared to normally fluent speakers.... Do people who stutter... self-report more speech errors? actually make more speech errors? – In inner speech? – In overt speech?
The current study Tonguetwister repetition 32 people who stutter 32 normally fluent controls - matched for age, gender and education 48 tonguetwisters per participant Speech-rate carefully controlled Dependent Variables – Onset errors Self reports Experimenter ratings – Word-order errors Self reports Experimenter ratings Masking (pink noise) No masking Inner speech 12 Out loud12
The current study Tonguetwister repetition 32 people who stutter 32 normally fluent controls - matched for age, gender and education 48 tonguetwisters per participant Speech-rate carefully controlled Dependent Variables – Onset errors Self reports Experimenter ratings – Word-order errors Self reports Experimenter ratings Masking (pink noise) No masking Inner speech 12 Out loud12
procedure Each participant recites forty eight, 4-word tonguetwisters e.g. Lean reed reef leach Each tonguetwister repeated 8 times to a (visual) metronome... – 4 x 1 word/sec – 4 x 2 words/sec
lean reed reef leach (familiarization phase)
lean reed reef leach.
lean reed reef leach -
lean reed reef leach --
lean reed reef leach ---
lean reed reef leach ----
lean reed reef leach.
lean reed reef leach -
lean reed reef leach --
lean reed reef leach ---
lean reed reef leach ----
lean reed reef leach.
lean reed reef leach -
lean reed reef leach --
lean reed reef leach ---
lean reed reef leach ----
lean reed reef leach.
lean reed reef leach -
lean reed reef leach --
lean reed reef leach ---
lean reed reef leach ----
lean reed reef leach Press SPACEBAR to continue
-
--
---
----
Type any errors in the space below Then press SPACEBAR to continue lean reed reef leach |
Press SPACEBAR to continue
-
--
---
----
Type any errors in the space below Then press SPACEBAR to continue Lean reed reef leach |
lean reed reef leach Press SPACEBAR to continue
-
--
---
----
Type any errors in the space below Then press SPACEBAR to continue Lean reed reef leach |
results onset errors – self-ratings e.g. lean reed reef leach→ Lean reed leaf leach PWS self-report more errors*** Overt errors more frequently self-reported* No significant interactions.
results word-order errors e.g. lean reed reef leach→ Lean reed leach reef PWS self-report more errors** Overt errors more frequently self-reported** No significant interactions
Compared to controls... the PWS group self-reported more errors, – both in inner and in overt speech However, did they actually make more errors? – how accurate were their self-reports?
results onset errors – monitoring vigilance e.g. lean reed reef leach→ Lean reed leaf leach Fewer self-reports than experimenter reports*** No significant interactions. For both PWS and controls... the ratio of self-reports to experimenter reports is similar
e.g. lean reed reef leach→ Lean reed leach reef Fewer self-reports than experimenter reports* No significant interactions For both PWS and controls... the ratio of self-reports to experimenter reports is similar results word-order errors – monitoring vigilance
PWS self-reported more errors than Controls Monitoring vigilance of PWS & Controls was similar Therefore we can conclude that.... PWS actually made more errors than Controls – in overt speech – and also in inner speech – Phonological encoding errors – and also Word-order errors
Is the severity of stuttering related to the number of inner-speech errors PWS self- report? – According to the Covert Repair Hypothesis, it should be. One final question...
Stuttering severity (SSI4) scores not correlated to onset errors self reports (inner speech) self reports (overt speech) independent rater reports SSI4 score onset errors vs. Stuttering Severity in PWS number of errors
onset errors vs. Fluency Difficulty in PWS and controls Fluency Difficulty self-ratings not correlated to speech errors
conclusions Compared to normally fluent speakers, PWS are less proficient at phonological encoding and make more phonological encoding errors. They are also less proficient at other aspects of language encoding. However, the tendency to make more errors of phonological encoding does not account for the severity of stuttering-like disfluencies as measured by the SSI4 or participants’ own self- ratings
The Edinburgh Disfluency Group References Oppenheim, G., & Dell, G. (2008). Inner speech slips exhibit lexical bias, but not the phonemic similarity effect. Cognition, 106(1), Postma, A., & Kolk, H. (1992). Error monitoring in people who stutter: evidence against auditory feedback defect theories. J Speech Hear Res, 35(5), Postma, A., & Kolk, H. (1993). The covert repair hypothesis: prearticulatory repair processes in normal and stuttered disfluencies. J Speech Hear Res, 36(3), Riley, G. (2009). Stuttering Severity Instrument for Children and Adults. (SSI-4) (4th ed). Austin TX: Pro-Ed